Sounds like a plan.
IN ADDITION to this, if we're mad on redundancy, we also ''could''
consider splitting the cluster into two clusters, with round robin DNS
spreading the traffic load between these clusters -- that way, if one
entire cluster were to fall over, the Wikipedia would still work.
HOWEVER:
* I'm not sure how likely it is that an entire cluster falls over.
* This would create absolutely MENTAL replication traffic.
(Okay, maybe that's not such a good idea then. Anyway.)
-- ropers [[en:User:Ropers]]
www.ropersonline.com
On 1 Oct 2004, at 18:06, Jimmy (Jimbo) Wales wrote:
I'm a little confused on the process whereby we
make decisions about
shell access. I'm thinking right now primarily of access for a couple
of people from the German e.V. to work on their website, which doesn't
yet exist, rather than thinking of shell access to the main cluster.
Related to this, it seems sensible to me that we should buy a special
server to host things that we want to be separate, like for example
the donation pages, so that if wikipedia is blown up for some reason,
people can still donate money. This would also permit us to give
shell accounts more freely to people with a specific need on some
auxiliary website.
--Jimbo
--
"La nèfle est un fruit." - first words of 50,000th article on
fr.wikipedia.org
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)wikimedia.org
http://mail.wikipedia.org/mailman/listinfo/wikitech-l